[Bugs] [Bug 1138970] file corruption during concurrent read/write

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1138970



--- Comment #17 from Raghavendra G <rgowdapp at redhat.com> ---
<Avati>

My impression looking at the backlog is that, you found that the corruption
symptoms disappeared by conducting an experiment where the order of xlators
were reversed. Based on this alone you are suggesting the change in xlator
order itself as a fix.

</Avati>

No, I arrived at the experiment by observing that file was never really
corrupted on the bricks. So, this pointed to some caching gone bad. I had
disabled io-cache, read-ahead and quick-read on client and was still able to
reproduce the issue. So, this left three components which might be doing
caching:

1. python libraries
2. client side vfs
3. vfs on brick

Since I had seen two writes being issued to same offset (which were never
supposed to when this bug is not hit), I suspected something gone wrong with
stat as seen by python and vfs. So, I suspected caching in 1 and 2 and tried
flipping the order of md-cache and write-behind.

But anyways, to ascertain the actual cause I need to find out some way to test
2 above.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=8mmR61A1Lk&a=cc_unsubscribe